WIRED published an investigation on November 2 reporting that about 400 interlinked apps and websites, presented as pet portrait makers, language-learning tools, AI fitness coaches and the like, load a different interface under certain conditions, letting users upload photos of other people to generate explicit images and videos. The material amounts to nonconsensual sexual deepfake abuse, and some of it involves violence.
The trail began with video ads on Facebook and Instagram. Researchers followed thousands of explicit ads back to apps in the Apple and Google app stores, where payment ran through Apple Pay and Google Pay.
The switch happens after review
According to the report, after each download the app sends the server details such as where the user came from, their language and a unique identifier, and the server decides whether to show the ordinary version or the explicit one. Arriving through one of the Meta ads and then downloading appears to be one of the triggers, though the exact rules are unclear.
Apple calls this tactic a "bait-and-switch": a developer swaps the content source after review has been passed. The reviewer sees a pet portrait app, while a paying user may see something entirely different.
Among the examples in the report, Viefur describes itself as a "pet portrait" app and sells 130 coins for about $4; DiffAI was created in December 2025 and has more than 100,000 downloads on Google Play.
The figures on scale come from several research groups:
- AI Forensics: in September, 27 related apps were still on Google Play with more than 350,000 downloads combined, many rated suitable for all ages; related explicit ads this year number in the tens of thousands; at its peak the operation registered about 30 websites a week
- Infoblox: identified nearly 500 linked domains, which the researchers say "run the same code and reference each other"
AI Forensics researcher Paul Bouchaud's assessment:
"This is an organized, commercially run operation: One operator behind hundreds of disposable apps and domains"
How the platforms responded
After WIRED supplied its findings, Apple removed 19 apps and said it has updated its rules to restrict such tools further. Google did not respond to a request for comment, but had already removed most of the apps before reporters got in touch.
After receiving the list of domains, Meta first confirmed more than 3,000 related ads, about two thirds of which have been taken down. It later said it had acted against 11,600 ad accounts, more than 4,700 business pages, 5,200 users and 15,000 pages, and had blocked 209 apps from advertising. The report says the ads have already prompted investigations by regulators in several countries. The US nonprofit Tech Transparency Project had also previously found deepfake ads depicting minors.
Clues pointing to Chinese-speaking developers
The part of the report that readers in the Chinese-speaking world cannot ignore is the identity of the operator. WIRED and the researchers rely on a set of circumstantial evidence: extensive Chinese-language comments in the code, hosting exclusively inside mainland China, ad videos carrying the watermark of a Chinese video-editing app, and contact details, admin login pages and some email addresses for the first registered site that point to a Hong Kong-registered company, KeLing Technology Limited.
These are investigative judgments, and WIRED's wording leaves room for doubt. The company did not respond to a request for comment, and related public information was altered after reporters asked about it. Meta says the company was banned from advertising in 2025. The company's name is spelled differently from Kling, the video model from Kuaishou, and the report does not suggest any connection between the two, so readers should not confuse them.
Danish outlet DR reported on part of this in September. So far the coverage describes no enforcement case opened by law enforcement and no systematic penalties at the store level, and no organization has given an updated figure for how many apps in the network are still online.
Sources: WIRED, AI Forensics, CocoLoop, Infoblox; platform enforcement figures follow the numbers Apple and Meta gave the media, and the operator's identity is an investigative judgment.